In vivo effects of the anti–interleukin-6 receptor inhibitor tocilizumab on the B cell compartment

Authors


Abstract

Objective

Interleukin-6 (IL-6) receptor inhibition by tocilizumab was recently licensed for the treatment of rheumatoid arthritis (RA). IL-6 induces in vitro differentiation of B cells into antibody-forming cells; however, the in vivo effects of IL-6 inhibition on the B cell compartment are currently not known. The purpose of this study was to examine this feature.

Methods

Sixteen patients with active RA were treated in an open-label study with tocilizumab (8 mg/kg every 4 weeks). Immunophenotyping was performed at baseline, week 12, and week 24.

Results

Memory B cell subsets declined significantly during tocilizumab therapy. Preswitch memory B cells decreased from a median of 19.6% to 12.3% at week 24 and postswitch memory B cells declined from a median of 18.6% to 15.0% at week 24 (P = 0.04). In parallel, CD19+IgA+ and CD19+IgG+ B cells decreased significantly. The proportion of IgA-expressing B cells fell from a median of 9.2% at baseline to 4.3% at week 12 and to 3.6% at week 24 (P = 0.01). IgG+ B cells declined from a median of 6.7% at baseline to 4.9% at week 12 (P = 0.007) and 2.8% at week 24 (P = 0.01). In parallel, serum levels of IgA and IgG were significantly diminished at week 24 (P < 0.05). There was a good correlation between relative and absolute numbers of IgA+ B cells with serum IgA at week 24.

Conclusion

Tocilizumab induced a significant reduction in the frequency of peripheral preswitch and postswitch memory B cells. In addition, the number of IgG+ and IgA+ B cells declined and correlated well with reduced serum immunoglobulin levels. The data indicate that IL-6 blockade affects the B cell hyperreactivity in RA patients.

Interleukin-6 (IL-6) is a pleiotropic cytokine that is produced by various cell types, such as T cells, B cells, monocytes, fibroblasts, osteoblasts, keratinocytes, endothelial cells, and some tumor cells (1). IL-6 has broad biologic activities in different target cells, including bone metabolism, acute-phase protein synthesis, and various components of the immune system. It is expressed in high concentrations at sites of inflammation and plays a central role in chronic inflammation (2).

Signal transduction by IL-6 is mediated by a unique IL-6 receptor system containing 2 functional proteins known as IL-6 receptor (IL-6R) and glycoprotein 130 (gp130) molecules. Binding of IL-6 to IL-6R results in the formation of the IL-6/IL-6R complex, which induces homodimerization of gp130, a membrane-bound protein involved in non–ligand-binding signal transduction, and initiates responses by activating transcription factor STAT-3 along with other signaling molecules (3). Soluble IL-6R can also bind IL-6 and form a complex with gp130 subunits on cells that do not express the transmembrane IL-6R, a process known as trans-signaling (1). While gp130 is ubiquitously expressed, the transmembrane IL-6R is predominantly found on hepatocytes, neutrophils, monocytes, macrophages, and some lymphocytes (4).

IL-6 was initially identified as B cell stimulatory factor 2, which is important for the development of antibody-producing plasma cells (5). Excessive activity of IL-6 causes polyclonal B cell activation, plasmacytosis, and B cell neoplasia. Such wide ranges of IL-6 activities are believed to constitute an important link between adaptive and innate immunity by mediating the T cell and B cell responses involved in autoimmunity (6). In vitro studies in IL-6–knockout mice have shown a marked reduction in B cell immune responses (7), particularly reductions in the levels of IgG1, IgG2a, and IgG3 upon immunization with a T cell–dependent antigen (8), suggesting that IL-6 is central for the induction and/or maintenance of plasma cells that produce these immunoglobulin subclasses. A recent study showed that antibody production is indirectly promoted by B cell helper capabilities of CD4+ T cells through increased IL-21 production upon IL-6 stimulation (9).

The role of B cells in the pathogenesis of rheumatoid arthritis (RA) has become more widely appreciated in the last couple of years (10). IL-6 has been shown to be up-regulated in the serum and synovium of RA patients. Enhanced serum levels of IL-6 and soluble IL-6 receptors have been correlated with disease activity (11, 12) and joint destruction (13). B cell differentiation and selection in the inflamed synovium, including the formation of ectopic follicular structures, is a key finding in RA (14). Therefore, the effect of IL-6 on B cells appears likely to be important but has not been thoroughly addressed under in vivo conditions in which patients are being treated with an agent that inhibits its effects.

Nevertheless, in addition to its effect on B cells, other mechanisms of IL-6 also contribute to the pathophysiology of RA. IL-6 induces osteoclast and osteoblast dysregulation, leading to accelerated bone resorption and reduced bone formation (15). Moreover, it has been shown that IL-6 stimulates the production of vascular endothelial growth factor in the synovium of RA patients, thus contributing to pannus formation (16). Recent studies of IL-6R inhibition using tocilizumab have shown good clinical responses in RA (3, 17) and other immune-mediated diseases (3).

Tocilizumab is a humanized monoclonal anti-human IL-6R antibody with specificity for the IL-6 receptor; it binds soluble and membrane-expressed IL-6R. Tocilizumab has recently been approved for the treatment of adult patients with moderately to severely active RA with an inadequate response to one or more disease-modifying antirheumatic drugs (DMARDs) or tumor necrosis factor (TNF) antagonists (17, 18). It is given once every 4 weeks as a single 60-minute intravenous infusion at a dose of 4–8 mg/kg and may be used as monotherapy in cases of incompatibility with methotrexate or other DMARD.

Apart from the clinical efficacy and safety of tocilizumab, there are actually no data available concerning the influence of therapeutic IL-6 inhibition on the B cell compartment in vivo. We therefore conducted a prospective trial in patients with RA and investigated the effects of blocking IL-6 with tocilizumab on different peripheral B cell subsets over a 24-week study period. Because of the known effects of IL-6 on the late stages of B cell differentiation in animal models, we were particularly interested in its effects on memory and class-switched B cells in vivo.

PATIENTS AND METHODS

Patient characteristics and study design.

Consecutive RA patients qualifying for anticytokine therapy were offered enrollment in a prospective open-label study of intravenous tocilizumab infusions. The primary end point was set at 12 weeks, with an extension period to 24 weeks. Sixteen RA patients (10 women, 6 men) were included in the trial. Informed consent was obtained from all patients prior to study entry, in accordance with the protocol approved by the Ethics Committee of the University of Würzburg. Tocilizumab was given with approval of the European Medicines Agency. A dose of 8 mg/kg was administered every 4 weeks as a 60-minute infusion.

All 16 RA patients met the American College of Rheumatology criteria for RA (19). Their mean disease duration was 11.7 years (range 2–44 years), and their median age was 54.7 years (range 30.4–75.7 years). The patients had failed to respond to treatment with standard DMARDs and/or TNFα antagonists. Eleven of the patients had had an inadequate response to MTX or leflunomide, and 5 patients had failed to respond to anti-TNF therapy (etanercept in 2, infliximab in 2, and adalimumab in 1). Thirteen of the 16 patients were receiving concomitant therapy with MTX, and the other 3 patients continued to take leflunomide. At study entry, 8 of the 16 patients were rheumatoid factor (RF) positive, and 6 of the 16 patients were anti–cyclic citrullinated peptide (anti-CCP) antibody positive. One of the RF-negative patients had been positive for RF in the past, but had become RF-negative during previous courses of therapy before entering the study.

In patients receiving anti-TNF agents, therapy was stopped a median of 1 month (range 1–6.1 months) prior to initiation of tocilizumab. Of the 5 patients who had taken anti-TNF agents, therapeutic failure was primary in 3 of them and secondary in the other 2.

At week 12, the Disease Activity Score in 28 joints (DAS28) (20) had declined significantly, to a mean ± SEM of 2.49 ± 0.4, from 4.9 ± 0.2 at baseline (P < 0.001). According to the response criteria of the European League Against Rheumatism (21), 11 patients were classified at week 12 as good responders, 5 as moderate responders, and 1 as a nonresponder. The serologic parameters we assessed (C-reactive protein [CRP] and erythrocyte sedimentation rate [ESR]) declined in parallel (Table 1). The baseline characteristics of the 16 study patients, along with the laboratory and clinical followup data, are shown in Table 1.

Table 1. Clinical and laboratory data at baseline and during therapy with tocilizumab in 16 patients with rheumatoid arthritis*
 BaselineWeek 4Week 12Week 24
  • *

    The primary end point of the study was a reduction in the Disease Activity Score in 28 joints (DAS28) at week 12. Except where indicated otherwise, values are the mean ± SEM. CRP = C-reactive protein; ESR = erythrocyte sedimentation rate; RF = rheumatoid factor; HAQ = Health Assessment Questionnaire; DI = disability index; VAS = visual analog scale (100-mm).

  • P < 0.001 versus baseline.

  • P < 0.05 versus baseline.

Age, median (range) years54.7 (30.4–75.7)   
% female62.5   
Disease duration, mean (range) years11.7 (2–44)   
DAS28 score4.9 ± 0.23.2 ± 0.42.49 ± 0.42.51 ± 0.4
CRP, mg/dl1.2 ± 0.40.07 ± 0.010.09 ± 0.020.2 ± 0.1
ESR, mm/hour19 ± 4.710 ± 45.3 ± 1.38.9 ± 3.2
RF, units/ml63.5 ± 14.653.6 ± 12.847.5 ± 11.145.9 ± 11.2
HAQ DI1.1 ± 0.10.8 ± 0.20.8 ± 0.20.9 ± 0.1
Global assessment of disease activity, by VAS    
 Patient's62.9 ± 5.753.4 ± 642.2 ± 6.538.4 ± 5.6
 Physician's46.2 ± 4.737.5 ± 3.524.7 ± 5.118.6 ± 1.8

Assessments.

Patients were assessed at baseline and at weeks 4, 12, and 24 for clinical and laboratory data. The primary end point of the study was a reduction in the DAS28 at week 12. Laboratory measurements included the ESR, the serum CRP concentration, serum levels of IgG, IgA, and IgM, RF levels, and anti-CCP antibody levels. Peripheral blood samples for immunophenotyping were obtained from all patients at baseline, week 12, and week 24. Immunophenotyping for CD19, CD27, IgD, CD38, CD20, and IgM was performed on all patients, using a whole blood method. In 10 patients, additional staining for CD19, CD27, IgD, CD38, CD10, CD20, IgA, and IgG coexpression was performed. For IgG or IgA surface staining, Ficoll-Paque separation was done in order to obtain reliable staining results.

Flow cytometric analysis.

Whole blood (200 μl) was used, and after washing twice with phosphate buffered saline (PBS) and albumin (2.5 gm/500 ml), samples were incubated for 15 minutes in PBS with 10 μl of monoclonal antibodies (mAb). Cells were then incubated for 10 minutes with 2 ml of VersaLyse (Beckman Coulter) and washed with PBS. Five-color staining was performed using an FC500 instrument (Beckman Coulter). Phenotype analyses were performed using CD19 (PC7), CD27 (phycoerythrin [PE]), CD20 (energy-coupled dye [ECD]), CD38 (PC5), anti-human IgD (fluorescein isothiocyanate [FITC]), and anti-human IgM (PC5). The following mAb were used as isotype controls: mouse IgG1/IgG2a (FITC/PE), mouse IgG1 (PC5), mouse IgG1 (ECD), and mouse IgG1 (PC7). All antibodies were from Beckman Coulter.

B cells were identified by forward scatter versus side scatter gating on viable lymphocytes in combination with gating on CD19+ cells. Ten thousand CD19 events were collected for each analysis. Frequencies of CD19+ cells were calculated using CXP software (Beckman Coulter).

In 10 of the 16 patients, peripheral blood mononuclear cells (PBMCs) were additionally prepared by separation with Ficoll-Paque Plus (Pharmacia Biotech). PBMCs were incubated in PBS with 10 μl of mAb for 20 minutes on ice. In each tube, 1 × 106 cells were suspended. The cells were then washed in PBS. Four-color staining was performed using a FACSCalibur instrument (Becton Dickinson). Phenotype analyses were performed using CD19 (allophycocyanin [APC]), CD27 (PE), anti-human IgD (FITC), CD10 (PE), CD38 (PerCP), CD20 (FITC), anti-human IgG (biotin–streptavidin–PerCP), and anti-goat IgA (biotin–streptavidin–PerCP). As isotype controls, the following antibodies from Becton Dickinson were used: mouse IgG1/IgG2a (FITC/PE), mouse IgG1 (PerCP–Cy5.5), and mouse IgG1 (APC). The total numbers of B cells of various phenotypes were calculated per microliter of blood, based on the frequencies of these cells among the lymphocytes and on the white blood cell count.

Controls.

Peripheral blood was obtained from 21 healthy volunteer donors (4 men and 17 women). Their mean age was 41 years (range 23–62 years). Flow cytometric analysis of different B lymphocyte populations was done using CD19, CD27, IgD, and IgM coexpression.

Statistical analysis.

Statistical analysis was performed with the nonparametric Wilcoxon test using GraphPad Prism software version 3.03. Values were always compared with baseline levels. For correlation of different B cell subsets between healthy controls and RA patients, the nonparametric Mann-Whitney test was used. P values less than 0.05 was considered significant. Linear regression was used to correlate clinical response with serum immunoglobulin levels and with different B cell subsets.

RESULTS

Comparison of peripheral B cell subsets in RA patients and healthy controls.

Before initiation of tocilizumab therapy, we compared different peripheral B cell subsets defined by the coexpression of CD19, CD27, and IgD in RA patients and 21 healthy controls. Notably, the RA patients had a significantly higher frequency of postswitch CD27+IgD− B cells before starting therapy with tocilizumab than did healthy controls (P = 0.03) (Figure 1A). Moreover, preswitch memory B cells were also increased in RA patients as compared to healthy controls, but this difference did not reach statistical significance (P = 0.06) (Figure 1B).

Figure 1.

Comparison of preswitch and postswitch memory B cells in rheumatoid arthritis (RA) patients (n = 16) and healthy controls (n = 21), and time course of changes in these populations during therapy with tocilizumab in RA patients. A, Frequency of CD27+IgD− postswitch memory B cells. There was a significantly higher frequency of this B cell population in RA patients at baseline than in the controls. After starting therapy with tocilizumab, this population decreased significantly. B, Frequency of CD27+IgD+ preswitch memory B cells. There was an increased frequency of this B cell population in RA patients at baseline than in the controls (P = 0.03). After starting therapy with tocilizumab, this population decreased significantly, both at week 12 and at week 24. Data are shown as box plots, representing the 10th, 25th, 50th (median), 75th, and 95th percentiles of the frequencies of total B cells.

Corresponding to the memory B cell populations, the percentages of CD27−IgD+ naive B cells were significantly lower in RA patients as compared to healthy controls (median 60.2% [range 18.8–78.5%] in RA patients versus 66.1% [range 38.6–84.2%] in controls; P = 0.04). Due to a significant absolute B cell lymphopenia (median of 86.4 cells/μl [range 26–261] in RA patients versus 229.5 cells/μl [range 111.3–333.5] in controls; P < 0.0001), the absolute numbers of preswitch memory B cells, postswitch memory B cells, and naive B cells were significantly lower in RA patients than in healthy controls: for CD27+IgD− B cells, a median of 14.1 cells/μl (range 3.7–70) in RA patients versus 37.8 cells/μl (range 11.2–88.3) in controls, for CD27+IgD+ B cells, a median of 18.0 cells/μl (range 3.7–175) in RA patients versus 38.6 cells/μl (range 15.5–100.1) in healthy controls, and for CD27−IgD+ B cells, 51.3 cells/μl (range 13.3–138.8) in RA patients versus 160.7 cells/μl (range 71.8–360.9) in healthy controls.

B cell homeostasis during tocilizumab therapy.

The relative numbers of total CD19+ B cells remained stable during therapy with tocilizumab (median 7.2% [range 3.3–14.8%] at baseline and 8.3% [range 2.9–16.7%] at week 24). However, the absolute numbers of CD19+ B cells increased from a median of 86.4 cells/μl (range 26–261) at baseline to 122.1 cells/μl (range 38.5–409.7) at week 24.

Subclass distribution of the memory B cell compartment.

A significant decline in the relative contribution of preswitch and postswitch memory B cells was observed during tocilizumab therapy at week 24. Preswitch memory B cells decreased significantly, from a median of 19.6% (range 3.4–39.0) at baseline to 13.3% (range 4.9–37.2) at week 12 (P = 0.02) and remained at reduced levels of 12.3% (range 5.5–38.1) at week 24 (P = 0.04). Postswitch memory B cells declined from a median of 18.6% (range 7.1–32.2) at baseline to 17.0% (range 7.4–32.9%) at week 12 and reached a significant decline to 15.0% (range 6.7–24.5) at week 24 (P = 0.03) (Figure 1).

The absolute numbers of CD27+IgD− B cells were reduced from 14.1 cells/μl (range 3.7–70) at baseline to 12.4 cells/μl (range 7.3–33.3) at week 24, and the absolute numbers of CD27+IgD+ B cells were reduced from 18.0 cells/μl (range 3.7–175) at baseline to 16.4 cells/μl (range 5.4–25.2) at week 24, but this difference did not reach statistical significance. CD27+CD38highIgD−CD20− plasmablasts showed no significant changes in relative and absolute numbers under tocilizumab therapy during the period of observation.

Analysis of immunoglobulin class–switched peripheral B cells.

In 10 of the 16 RA patients, additional immunophenotyping was performed, with staining for the immunoglobulin isotypes IgA and IgG and coexpression of CD19, CD27, and CD20. Here we observed significant decreases in the relative and absolute numbers of total IgA+ and IgG+ B cells at week 24 under therapy with tocilizumab (Figure 2). Of interest, IgA+ B cells were significantly reduced, from a median of 9.2% (range 3.2–14.6) at baseline to 3.6% (range 0.8–8.1) at week 24 (P = 0.01). In addition, the absolute numbers of IgA+ B cells showed a significant decrease, from a median of 13.3 cells/μl (range 6.8–22.4) at baseline to 7.5 cells/μl (range 0.6–22.7) at week 12 (P = 0.007), and they then declined further to 4.9 cells/μl (range 1.1–12.5) at week 24 (P = 0.007).

Figure 2.

Time course of changes in the frequencies of IgA+ and IgG+ peripheral B cells during therapy with tocilizumab in rheumatoid arthritis patients (n = 10). Both the relative numbers (A) and the absolute numbers (B) of IgA+ B cells declined significantly after starting therapy with tocilizumab. In addition, both the relative numbers (C) and the absolute numbers (D) of IgG+ B cells declined during tocilizumab therapy. Data are shown as box plots, representing the 10th, 25th, 50th (median), 75th, and 95th percentiles of the frequencies of total B cells.

IgG+ B cells showed similar results, with a significant reduction in the frequencies at week 12 (P = 0.007) (Figure 2C). The absolute numbers of these cells were also decreased from a median of 8.2 cells/μl (range 2.9–27.5) at baseline to 3.6 cells/μl (range 1.9–14.5) at week 24 (P = 0.035) (Figure 2D).

Since IL-6 has been shown in animal models to be important for plasma cell differentiation and survival (22, 23), further analyses were performed to determine serum levels of IgG, IgA, and IgM at weeks 4, 12, and 24. Of note, levels of serum IgG and IgA decreased significantly at week 12 and week 24 as compared to baseline (Figure 3). Serum levels of IgM remained stable during therapy with tocilizumab.

Figure 3.

Time course of changes in serum levels of IgG, IgA, and IgM during therapy with tocilizumab in rheumatoid arthritis patients (n = 16). After starting therapy with tocilizumab, serum IgG and IgA levels declined significantly at week 12 and week 24 compared with baseline (∗ = P < 0.05 and ∗∗ = P = 0.007). Values are the mean ± SEM.

Naive B cell compartment under IL-6 inhibition.

The absolute numbers of naive CD27−IgD+ B cells increased from a median of 51.3 cells/μl (range 13.3–138.8) at baseline to 67.4 cells/μl (range 14.7–214.2) at week 24 (P = 0.06), while the frequency of naive CD27−IgD+ B cells did not change during the observation period. Interestingly, transitional-type CD38highIgD+CD10+ B cells showed a significant increase in relative numbers at week 24 as compared to baseline (P = 0.005) (Figure 4A). Similar changes were found in the absolute numbers, with a significant increase at week 24 as compared to baseline (P = 0.002) (Figure 4B).

Figure 4.

Time course of changes in transitional-type CD38highIgD+CD10+ B cells during therapy with tocilizumab in rheumatoid arthritis patients (n = 10). Both the relative numbers (A) and the absolute numbers (B) of CD38highIgD+CD10+ B cells increased during therapy with tocilizumab and were significantly increased by week 24 as compared with baseline. Data are shown as box plots, representing the 10th, 25th, 50th (median), 75th, and 95th percentiles of the frequencies of total B cells.

Relationship between clinical parameters and B cell subsets.

We further examined a possible relationship between clinical parameters and changes in different B cell subsets. There was a significant correlation between IgA+ B cells and serum levels of IgA. The relative and absolute numbers of IgA+ B cells correlated well with serum IgA levels at week 24 (P = 0.0016 and P = 0.03, respectively) (Figures 5A and B). By analyzing the clinical response according to a reduction in the DAS28 score, we found a correlation between changes in the DAS28 scores and the relative numbers of preswitch memory B cells (Figure 5C). The other subpopulations were not significantly correlated.

Figure 5.

Linear regression analyses of clinical parameters and B cell subsets in patients with rheumatoid arthritis (RA). A and B, Correlation of serum levels of IgA with the absolute (A) and relative (B) numbers of CD19+IgA+ B cells at week 24 after starting therapy with tocilizumab in 10 RA patients. Both the absolute and relative numbers of IgA+ B cells correlated well with reduced serum levels of IgA. C, Correlation of the Disease Activity Score in 28 joints (DAS28) with the relative numbers of CD27+IgD+ preswitch memory B cells at week 12 after starting therapy with tocilizumab in 15 RA patients. There was a correlation between reductions in the DAS28 scores and reductions in the relative numbers of preswitch memory B cells.

DISCUSSION

Interleukin-6 is one of the key cytokines involved in the pathophysiology of RA. An increased level of IL-6 in different autoimmune diseases indicates that IL-6 and IL-6–mediated signaling cascades are potential targets for autoimmune therapy. IL-6 receptor inhibition using tocilizumab has been documented in various clinical trials to reduce signs and symptoms (3, 17), as well as radiologic progression, in patients with RA (24) and other inflammatory diseases (25, 26).

The exact mechanism of IL-6 inhibition that is responsible for the clinical response has not been fully elucidated. In mouse models, IL-6 is reported to be involved in B cell hyperactivity, autoantibody production, and immunopathology (27). In RA patients, chronic activation of B cells and an accumulation of memory B cells in the peripheral blood and the synovial membrane have been described (28, 29). Within this context, B cell–targeted therapies have been widely explored in RA (30). Since IL-6 has been described as an important B cell differentiation factor with effects on plasma cell differentiation and survival in the bone marrow (22), the aim of the current study was to investigate the influence of the anti–IL-6 monoclonal antibody tocilizumab on the homeostasis of the peripheral B cell compartment as well as its effects on blood immunoglobulin levels in patients with RA.

The distribution of memory B cell subsets in the peripheral blood of our study cohort revealed a significantly higher frequency of postswitch memory B cells in the RA patients than in the healthy controls (Figure 1). Similarly, the proportion of preswitch memory B cells was higher in RA patients prior to tocilizumab treatment than in healthy controls. These results have some limitations, since they may reflect the specific composition of our patient cohort with regard to disease duration, anti-CCP status, or prior treatment with anti-TNF agents, all of which have been indicated to influence B cell homeostasis. Nevertheless, other studies have also reported a significant increase in overall peripheral memory B cells in RA patients (28), particularly the postswitch memory B cell subset (29).

The contribution of preswitch memory B cells in RA seems to be the most controversial feature. The levels have been reported to be comparable to those in healthy controls (29), but other investigators found a lower frequency of these cells in the periphery (31), particularly in patients with very early RA and a disease duration of less than 6 months (32). Also, anti-TNFα therapy seems to be related to increased numbers of preswitch memory B cells (31, 33). Due to the limited number of patients in our cohort, we were not able to perform valid subgroup analyses for this. Nevertheless, patients in our cohort who had a short disease duration (median <4 years) had also a trend toward lower percentages of preswitch memory B cells, as did our patients in whom DMARD treatment had failed, but were still higher than the percentages in the control group (data not shown).

Overall, our findings are consistent with the model of dynamic alterations of memory B cell subset composition during the course of disease, with accumulation of memory B cells in the periphery during evolution of RA, indicating enhanced B cell activation in patients with RA.

We therefore analyzed the impact of continuous IL-6 inhibition using the anti–IL-6R antagonist tocilizumab on the homeostasis of peripheral B cell subsets. Of central interest, the main impact of IL-6R blockade could be detected within the memory B cell compartment. We show a significant decrease in the frequency of CD27+IgD− postswitch memory B cells and CD27+IgD+ preswitch memory B cells in vivo under therapeutic IL-6 inhibition. The absolute numbers of both subpopulations showed no significant decrease; this was likely due to the parallel increase in the absolute numbers of B cells under therapy, as indicated above. The reduction in the frequency of preswitch memory B cells was detected as early as week 12 and for the postswitch subset after 6 months. The relatively early reduction in preswitch memory B cells may reflect their turnover rate, since studies of splenectomized and congenital asplenic patients (34, 35) indicate a shorter lifespan of preswitch than postswitch memory B cells (34). Preswitch memory B cells have been related to circulating marginal zone B cells. In studies of temporal B cell depletion using rituximab, the extent of repopulation of B cells has been linked to the clinical response in RA (36). Interestingly, we also found a correlation between decreasing preswitch memory B cells and the clinical response (Figure 5C).

IL-6 acts on different targets in the pathophysiology of RA. Currently, it can only be hypothesized if the effects on B cells described herein contribute to the mechanism of action of tocilizumab in RA. In systemic lupus erythematosus (SLE), expanded CD38highCD19lowIgD– plasmablasts were seen in patients experiencing a flare and were reported to be reduced under IL-6 inhibition (37, 38). In RA, plasmablast recirculation is not characteristically elevated, and therefore, no changes were identified in our study. However, IL-6 inhibition resulted in changes within the naive B cell compartment. In our study, we identified an increase in the proportion and absolute numbers of CD38highIgD+CD10+ transitional-type B cells, indicating that under IL-6 inhibition, the regeneration capacity of the B cell compartment was enhanced, possibly contributing to a juvenilization of peripheral B cells, and reducing the burden of memory B cells, which may also carry memory for the disease.

The further analysis of peripheral class-switched IgA+ and IgG+ B cells is consistent with that hypothesis. We see for both subsets a significantly decreased frequency at week 24 under IL-6 inhibition in vivo. In addition, the decline in absolute numbers of IgA+ and IgG + B cells also reached statistical significance at week 24. In parallel, serum IgA and IgG, but not IgM, levels declined under IL-6 inhibition, and a good correlation between IgA+ B cells and serum IgA levels was found (Figure 5). It is important to note that all serum immunoglobulin levels stayed within in the normal range and did not drop below the lower limit of normal. These results are consistent with in vitro studies in which IL-6 was shown to be an important B cell differentiation factor for the development of immunoglobulin-secreting cells (5, 39). Culturing B cells in vitro with anti–IL-6 antibodies inhibited immunoglobulin production even when added on the fourth day of an 8-day culture, indicating that IL-6 acts on the late phase of the B cell response (5). Interestingly, a marked reduction of mucosal IgA–producing cells has been observed, along with deficient local antibody responses in IL-6–deficient mice (40). In vivo data from an open-label study of tocilizumab in SLE indicate a reduction in anti-DNA antibody and serum IgG levels (37). In our study, RF levels did not decline significantly. This may simply be due to the proportion of RF-negative patients in our cohort. In future studies, however, the influence of tocilizumab on IgA-RF may be of particular interest.

The results of this study substantiate a clinically meaningful effect of systemic IL-6 blockade on B cell differentiation, rather than resembling only changes in the peripheral recirculation of memory B cell subsets. The reduced immunoglobulin levels have not yet been reported to be correlated with higher rates of infection. However, further studies of patients undergoing IL-6 inhibition that include analyses of specific protective immunoglobulin titers are needed.

In conclusion, we found a significant reduction in the frequency of preswitch and postswitch memory B cells, along with IgA+ and IgG+ B cells, under therapeutic IL-6 receptor inhibition. The initially higher frequencies of memory B cells in the periphery of our RA patients were normalized under IL-6 inhibition. The reduced frequency of preswitch memory B cells was associated with a reduction in the DAS28 under tocilizumab treatment. In parallel with the inhibition of the memory B cell compartment, serum levels of IgA and IgG were also diminished under treatment, with a significant correlation of the relative and absolute numbers of peripheral IgA+ B cells with the serum IgA levels. Our results establish the effects of IL-6 inhibition on the B cell compartment in vivo and are consistent with a therapeutic impact on B cell hyperactivity in patients with RA.

AUTHOR CONTRIBUTIONS

All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Tony had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study conception and design. Roll, Einsele, Dörner, Tony.

Acquisition of data. Roll, Muhammad, Schumann, Kleinert.

Analysis and interpretation of data. Roll, Muhammad, Schumann, Kleinert, Tony.

Acknowledgements

The authors thank Anette Koss-Kinzinger and Isabelle Kuntzsch for technical assistance.

Ancillary